torch_harmonics.distributed.DistributedResampleS2#

class torch_harmonics.distributed.DistributedResampleS2(
nlat_in,
nlon_in,
nlat_out,
nlon_out,
grid_in='equiangular',
grid_out='equiangular',
mode='bilinear',
)[source]#

Bases: Module

Distributed resampling module for spherical data on the 2-sphere.

This module performs distributed resampling of spherical data across multiple processes, supporting both upscaling and downscaling operations. The data is distributed across polar and azimuthal directions, and the module handles the necessary communication and interpolation operations.

See also

torch_harmonics.ResampleS2

Serial counterpart with full mathematical description and parameter documentation.

Parameters:

  • nlat_in (int) – Number of input latitude points

  • nlon_in (int) – Number of input longitude points

  • nlat_out (int) – Number of output latitude points

  • nlon_out (int) – Number of output longitude points

  • grid_in (str, optional) – Input grid type, by default "equiangular"

  • grid_out (str, optional) – Output grid type, by default "equiangular"

  • mode (str, optional) – Interpolation mode ("bilinear" or "bilinear-spherical"), by default "bilinear"

On this page